1(x,y,ñ) = ABs Dan) |
Fig. 2 shows the intensity of structure information on the
same profile as Fig. 1. Theoretically, we can get infinite
number of structure information intensity images. Actu-
ally, the synthesis of finite number of such images will
approximate the structure information on original image.
The synthesis algorithm is
IC) = 3 [Hx,y,n)] n € [0°,360°)
For example, we use this method to process the MSS
image Oct. , 1978) of Liangcheng district, Inner Mon-
golia. Eighteen directional structure images (I1,I2, ...,
I18, corresponding to eighteen directions with degree
space of 20?) were calculated , and synthesized , see Fig. 3 .
In Fig. 3, each pixel has a value, which expresses the
synthetic intensity of structure development at that point.
The result proves that almost all structure and texture in-
formation was extracted and expressed on the final image.
This result image (Fig. 3) has been successfully integrated
with other geophysical data to evaluate the potential re-
sources in this area.
'This method can only be used to operation on grid images.
For vector maps, representing geological structure lines,
the following method will solve the same problem.
2. 2 Techniques of Mathematical Morphology
(TMM)
This method was first used by J. serra? (1982) to
petrology research under microscope. But, up till now, it
is still not used to image processing. The main idea of
TMM is, (1) adjacent spatial targets are related to one
another, and more than one related targets may form a
patterns (2) spatial informationmay be extracted by
changing or reconstructing the spatial pattern of image or
any given target. Dilation is one of the basic operations of
TMM to extract spatial information. Dilation, as it
means, is to make the target dilate under specific rules to
its adjacent area.
For a vector binary map, convert it to grid image P(x,
y). which has only two logical values (0,1)
1, (x,y) is on targets
P(x,y) = :
0, (x,y) is not on any targets
Employ the following 3X 3 pallet to detect the concatena-
tion of position C,
X; X; X
X. 0 X
Xs X; Xs
loneof Xi G=1, 2, ... » 8) is logical 1, then Ç is a
concatenate point and be assigned the value of 1. The
typical dilation operation is as follows,
466
0 0 0 0 0 0 © 0 0 0
0. 0-0 > 0:0. 07 0.1
06 0 | 0 1 1 1 0 jf 1 9
0 0 9 0 0 0 0 0 0-0 9
10 0 -—1 4 TI 0 0--1 1 0
1 1 0 1 1 1 1 1 à 1 T
Q0. 4 0.90 1 0 0 1 0.0.1
0:1 ..=>0 1 1 1-041 = 1.1 1
11, 1 1 1 1 I 1 1 T 1 1
1 1 9 I 1 0 1 1 1 hil 1
i 01 --1 1 1 10 1-1 1 |
rir 111 1-1 1 I 1
By the above operations, all targets will be dilated one
times. Theoretically, the targets may fulfill the whole
space by recursive dilation. But that is not necessary.
Fig. 4.1 and Fig. 4.2 is the original grid structure map
and its corresponding eight times dilated image. You can
see that the structure pattern may be recognized obvious-
ly.
For geological usage, dilation operation will be more com-
plicated, because structure information cannot be ex-
pressed by only two logical values (0,1) ,and the distance
to the structure center is an important factor representing
of the relation between structure and mineral resources.
That is, the spatial structure information IS(x,y) is given
by
IS x.y} = [(D) s (1)
where D = di + d2
and, dl, d2 are the distance from (x, y) to the two
nearest structure centers,as illustrated in Fig. 5. More di
(1272) may add to D for different work area. Generally,
d1 + d2 will give satisfactory results. After practical ap-
plications, function f in (1) has been found to be expo-
nential as follows
V*kFXPk*D), DS DO ...... (2)
0, D> Do
IS =
where
V; attribute values on structure line
k: a negative coefficient dependent of nature of
different working area
DO: an arbitrary threshold distance beyond which no
structure information exists
IS will be the value of the corresponding dilated position (
signed C at pallet center above).
In Liangcheng, Inner Mongolia, the geological structure
an
(F
ag
ar
Li
TI
MC
Tt
tui
thi
en
inc
pre
Pi
De
IE
[1
rat
rel
[2
Sol
Nc
[3
us!
ant
